Ship collisions often result in huge losses of life, cargo and ships, as well as serious pollution of the water environment. Meanwhile, it is estimated that between 75% and 86% of maritime accidents are related to human factors. Thus, it is necessary to enhance the intelligence of ships to partially or fully replace the traditional piloting mode and eventually achieve autonomous collision avoidance to reduce the influence of human factors. In this paper, we propose a multi-ship automatic collision avoidance method based on a double deep Q network (DDQN) with prioritized experience replay. Firstly, we vectorize the predicted hazardous areas as the observation states of the agent so that similar ship encounter scenarios can be clustered and the input dimension of the neural network can be fixed. The reward function is designed based on the International Regulations for Preventing Collision at Sea (COLREGs) and human experience. Different from the architecture of previous collision avoidance methods based on deep reinforcement learning (DRL), in this paper, the interaction between the agent and the environment occurs only in the collision avoidance decision-making phase, which greatly reduces the number of state transitions in the Markov decision process (MDP). The prioritized experience replay method is also used to make the model converge more quickly. Finally, 19 single-vessel collision avoidance scenarios were constructed based on the encounter situations classified by the COLREGs, which were arranged and combined as the training set for the agent. The effectiveness of the proposed method in close-quarters situation was verified using the Imazu problem. The simulation results show that the method can achieve multi-ship collision avoidance in crowded waters, and the decisions generated by this method conform to the COLREGs and are close to the level of human ship handling.

Deep Reinforcement Learning (DRL) is widely used in path planning with its powerful neural network fitting ability and learning ability. However, existing DRL-based methods use discrete action space and do not consider the impact of historical state information, resulting in the algorithm not being able to learn the optimal strategy to plan the path, and the planned path has arcs or too many corners, which does not meet the actual sailing requirements of the ship. In this paper, an optimized path planning method for coastal ships based on improved Deep Deterministic Policy Gradient (DDPG) and Douglas–Peucker (DP) algorithm is proposed. Firstly, Long Short-Term Memory (LSTM) is used to improve the network structure of DDPG, which uses the historical state information to approximate the current environmental state information, so that the predicted action is more accurate. On the other hand, the traditional reward function of DDPG may lead to low learning efficiency and convergence speed of the model. Hence, this paper improves the reward principle of traditional DDPG through the mainline reward function and auxiliary reward function, which not only helps to plan a better path for ship but also improves the convergence speed of the model. Secondly, aiming at the problem that too many turning points exist in the above-planned path which may increase the navigation risk, an improved DP algorithm is proposed to further optimize the planned path to make the final path more safe and economical. Finally, simulation experiments are carried out to verify the proposed method from the aspects of plan planning effect and convergence trend. Results show that the proposed method can plan safe and economic navigation paths and has good stability and convergence.

Large ships are typically with large inertia and longtime delay in motion, in prevailing collision avoidance methods, their maneuverability is generally neglected, there could be a dangerous situation if the system fails to control the ship course as ordered in a timely manner. This paper proposes a coordination system which consists of two algorithms for avoiding risk and then returning to scheduled waypoint. The avoiding risk algorithm are based on VO (velocity obstacle) method, the returning algorithm is derived from LOS (light of sight) guidance. For better performance, the ship model for simulation is a nonlinear Norrbin Model, with the controller improved by CGSA (closed loop gain shaping algorithm) method from traditional PID control, COLREGS (Convention on the International Regulations for Preventing Collisions at Sea) constrains are considered. To test the effectiveness of the proposed system, a series of complex scenarios including Imazu problem are applied.

Merchant ships sometimes fail to detect small ships at night and in poor visibility, leading to urgent situations and even collisions. Infrared (IR) cameras have inherent advantages in small target detection and become essential environmental awareness equipment on unmanned ships. The existing target detection models are complex and difficult to deploy on small devices. Lightweight detection algorithms are needed with the increase in the number of shipborne cameras. Therefore, herein, a lightweight model for small IR ship detection was selected as the research object. IR videos were collected in the Bohai Strait, the image sampling interval was calculated, and an IR dataset of small ships was constructed. Based on the analysis of the characteristics of the IR ship images, gamma transform was used to preprocess the images, which increased the gray difference between the target and background. The backbone of YOLOv5 was replaced with that of Mobilev3 to improve the computing efficiency. Finally, the results showed that the parameters of the proposed model were reduced by 83% compared with those of the YOLOv5m model, while the detection performance was almost the same.